Convert NYC 2020-2021 School Year Calendar to csv for import to Google Calendar

calendar_events.sh

curl https://www.schools.nyc.gov/about-us/news/2020-2021-school-year-calendar > raw.html

cat raw.html | sed -z -e " \
    s/[\r\n]//g; 
    s/.*<table>//; 
    s/<\/table>.*//; 
    s/ </</g; 
    s/<\/*strong>//g; 
    s/<\/tr>\(<\(\/thead\|\/tbody\|tbody\|tr\)>\)*/\n/g; 
    s/<\/t[hd]><t[hd]>/\t/g; 
    s/<[^>]*>//g; 
    s/\&ndash;/-/g; 
    s/\&nbsp;/ /g; 
    s/\&rsquo;/'/g; 
    s/\&shy;//g; 
  " > calendar.txt

python calendar_filter.py calendar.txt > calendar.csv

calendar_filter.py

"""
calendar_filter.py
Converts tab-separated calendar data to csv for import to Google Calendar
"""
import fileinput

def convert_date(datestring, increment_day=False):
    if '-' in datestring:
        [start, end] = datestring.split('-')
        if ' ' not in end:
            [startmonth, startdom] = start.split(' ')
            end = startmonth + ' ' + end
        return tuple([convert_date(start)[0], convert_date(end, True)[0]])
    elif ' ' in datestring:
        [month, dom] = datestring.split(' ')
        if increment_day:
            # workaround for Google Calendar quirk: multi-day events imported with end shortened by one day
            dom = str(int(dom) + 1)
        mm = {'September': '09', 'October': '10', 'November': '11', 'December': '12',
                'January': '01', 'February': '02', 'March': '03',
                'April': '04', 'May': '05', 'June': '06', }[month]
        yyyy = '2020' if mm in ['09', '10', '11', '12'] else '2021'
        mmddyyyy = mm + '/' + dom + '/' + yyyy
        return tuple([mmddyyyy, mmddyyyy])
    else:
        return tuple(['Start Date', 'End Date'])


for line in fileinput.input():
    [date, dow, subject] = line.rstrip().split("\t");
    [start_date, end_date] = convert_date(date)
    if "NOTES" == subject:
        subject = "Subject"
        all_day_event = "All Day Event"
        description = "Description"
    else:
        subject = '"' + subject + '"'
        all_day_event = "True"
        description = '"https://www.schools.nyc.gov/calendar\nhttps://www.schools.nyc.gov/about-us/news/2020-2021-school-year-calendar"'

    print(",".join([subject, start_date, end_date, all_day_event, description]))